HPE EVA Storage

Continuous Access Question

 
SOLVED
Go to solution
Eric Datwyler
Frequent Advisor

Continuous Access Question

I am beginning the setup of DR with Continuous Access, two EVA 4400s, and Windows 2008 R2 servers. I am looking for some explanation of the "destination host access" setting as the documentation is very vague. When would I use "None", or "Inquire Only", or "Read Only"? It appears "Read Only" would be the best choice but why would you use one over the other? What are the Pros and Cons of each one?

Should a failure occur with one EVA, I need to know the setting that would provide the "quickest" recovery on the backup EVA. The source and destination servers will mostly be Microsoft Clusers (MSCS).
9 REPLIES 9
Azadeh
Advisor

Re: Continuous Access Question

Hi Eric

HP StorageWorks Cluster Extension (CLX) software offers protection against system downtime from fault, failure and disasters after using Replication in EVA

http://h20338.www2.hp.com/ActiveAnswers/cache/414462-0-0-0-121.html
Regards
Eric Datwyler
Frequent Advisor

Re: Continuous Access Question

Right, thanks but that doesn't have anything to do with my question. I am evaluating the cluster extension software, but for now, I need to know what the three settings will do for me without cluster extension.
Steven Clementi
Honored Contributor

Re: Continuous Access Question

Last I checked, I thought the default presentation setting was to mask the destination lun no matter what setting you had (RO, IO, etc).

In other words, if you present your destination vdisks to your servers, they won't be able to see the disks anyway until you failover. If so, the setting dos not matter when you are dealing with the servers that are accessing the source luns... you want them to be able to read/write at that time when/after you need to failover.


Steven
Steven Clementi
HP Master ASE, Storage, Servers, and Clustering
MCSE (NT 4.0, W2K, W2K3)
VCP (ESX2, Vi3, vSphere4, vSphere5, vSphere 6.x)
RHCE
NPP3 (Nutanix Platform Professional)
Eric Datwyler
Frequent Advisor

Re: Continuous Access Question

Right, but the problem is I need the "backup" servers ready and waiting with minimal time to bring up. Plus, MSCS needs to see the LUN to even add it as a disk resource.

I guess my core question is.....Is there any reason not to create them all as Read Only instead of None or Inquire Only? I just don't see any reason to have it as "None" or even "Inquire Only".
CLEB
Valued Contributor

Re: Continuous Access Question

The settings take effect when you failover the DR group.

This is what the help says:
#
None. The virtual disk cannot be presented to hosts.
#
Inquiry only. The virtual disk can be presented to hosts, but hosts can only make SCSI inquiries. No host I/O is allowed. This mode is typically used with host clusters.
#
Read only. The virtual disk can be presented to hosts, but for read only.


We leave ours set to None, which doesn't make sense as when we failover the LUN IS presented to the DR server. Then the servers boot automatically when they see the storage as they are powered on cycling the boot devices.
Uwe Zessin
Honored Contributor

Re: Continuous Access Question

Use "none" unless *YOU* can explain why you need the setting, how it works and how a file system works.

Sorry, but "backup" or "cluster failovers without CLX" are *NOT* valid reasons to use R/O or INQ_O.

In the case of R/O you risk writing corrupted data to tape or crashing your backup server -

no offense, but that's why I wrote: "explain ... how a file system works".
.
Eric Datwyler
Frequent Advisor

Re: Continuous Access Question

OK, since "backup" or "cluster failovers without CLX" are *NOT* valid reasons to use R/O or INQ_O......then what ARE valid reasons?

Also, I don't understand your comment about corrupt data to tape and crashing backup servers. Tape has nothing to do with anything and how would our backup servers crash when using Read Only as opposed to None? The data is still replicated in the same way.

Thanks to all. I hope I'm not sounding critical of everyone but I still haven't heard a valid reason to use one over the other in terms of Read Only, Inq Only, and None.
Uwe Zessin
Honored Contributor
Solution

Re: Continuous Access Question

R/O is used as an in-band communications channel for the CLX software in case the LAN is unavailable.
INQ_O is used in some special VMware failover environments to pre-populate SCSI targets/LUNS.

If you mount a destination vdisk R/O (if the operation system allows this...) while the EVA is going on with the replication in the background, the data/meta-data is changing all the time. Such changes are not expected by filesystems like NTFS which do local caching because they expect that no other instance is doing changes.

Inconsistencies in the meta-data can lead to crashes of the filesystem software which downs the entire OS. A 'stable' view to the filesystem is provided through snapshots or clones.
.
Eric Datwyler
Frequent Advisor

Re: Continuous Access Question

Excellent. That's the best information I've received/found yet. Thank You very much. I can now move forward with "None" as my option.